Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A point of sale system with facial recognition, the point of sale system comprising: one or more cameras; a processor device and memory coupled to the processor device, the processing system programmed to: receive a plurality of images from the one or more cameras; extract, with a feature extractor utilizing a convolutional neural network (CNN) with an enlarged intra-class variance of long-tail classes, feature vectors from each of the plurality of images; generate, with a feature generator, discriminative feature vectors for each of the feature vectors; classify, with a fully connected classifier, an identity from the discriminative feature vectors; and control an operation of the point of sale system to react in accordance with the identity.
A point of sale (POS) system with facial recognition is designed to enhance security and streamline transactions by identifying individuals using facial features. The system addresses challenges in accurately recognizing faces, particularly in scenarios with long-tail classes where some identities are underrepresented, leading to biased or inaccurate recognition. The system includes one or more cameras to capture images of individuals approaching the POS. A processor and memory are used to process these images. A feature extractor, utilizing a convolutional neural network (CNN) with an enlarged intra-class variance for long-tail classes, extracts feature vectors from each image. These feature vectors are refined by a feature generator to produce discriminative feature vectors, which are then classified by a fully connected classifier to determine the individual's identity. Based on the recognized identity, the POS system adjusts its operations, such as authorizing transactions, personalizing services, or triggering security measures. The system improves recognition accuracy and adaptability, especially in diverse or imbalanced datasets, ensuring reliable identity verification for secure and efficient transactions.
2. The point of sale system as recited in claim 1 , further includes a communication system.
A point of sale (POS) system is designed to process transactions at retail locations, but traditional systems often lack integrated communication capabilities, leading to inefficiencies in customer service and operational coordination. This invention enhances a POS system by incorporating a communication system to facilitate real-time interactions between staff, customers, and external entities. The communication system enables features such as instant messaging, notifications, and alerts, improving coordination among employees and reducing wait times. It also supports customer engagement by allowing direct communication between customers and staff, such as order status updates or assistance requests. Additionally, the system can integrate with external services like delivery tracking or inventory management to streamline operations. The communication system may use wired or wireless networks, including Wi-Fi, Bluetooth, or cellular connections, to ensure reliable and secure data transmission. By embedding communication directly into the POS infrastructure, the system enhances efficiency, customer satisfaction, and operational flexibility in retail environments.
3. The point of sale system as recited in claim 1 , wherein the operation logs a customer into the point of sale system and greets the customer.
A point of sale (POS) system is designed to facilitate transactions and customer interactions in retail or service environments. A specific implementation of this system includes a feature that automatically logs a customer into the system and provides a personalized greeting upon their arrival. This functionality enhances the customer experience by streamlining the checkout process and creating a more engaging interaction. The system may use identification methods such as biometric recognition, customer loyalty cards, or mobile device pairing to authenticate the customer. Once logged in, the system retrieves the customer's profile, transaction history, and preferences to tailor the greeting and subsequent interactions. The greeting may be delivered through a display screen, audio output, or a mobile application, depending on the system configuration. This feature reduces manual input requirements, speeds up transactions, and fosters customer loyalty by recognizing repeat visitors. The system may also integrate with inventory management and payment processing modules to provide a seamless experience from greeting to checkout.
4. The point of sale system as recited in claim 1 , wherein the operation logs an employee into the point of sale system and greets the employee.
A point of sale (POS) system is designed to streamline retail transactions by automating employee login and greeting processes. The system includes a user interface for employees to interact with the POS, a processing unit to execute transactions, and a data storage component to retain transaction records. The system also features a biometric authentication module, such as a fingerprint scanner or facial recognition, to verify employee identity before granting access. Once authenticated, the system logs the employee into the POS and displays a personalized greeting, such as "Welcome, [Employee Name]." This automation reduces manual login steps, enhances security, and improves the employee experience by providing a seamless and personalized interaction. The system may also track login times and employee activity for operational efficiency and compliance purposes. The biometric authentication ensures secure access while minimizing the risk of unauthorized use. The greeting feature enhances user engagement and reinforces a professional work environment. This system is particularly useful in high-volume retail environments where quick and secure employee access is critical.
5. The point of sale system as recited in claim 1 , wherein the operation recognizes a customer and permits a purchase without an employee intervention.
A point of sale (POS) system is designed to automate customer transactions in retail environments. The system includes a recognition mechanism that identifies customers without requiring manual input from employees. This allows purchases to be completed autonomously, eliminating the need for staff intervention during the transaction process. The system may use biometric data, digital payment methods, or other identification techniques to verify the customer's identity and authorize the purchase. By automating recognition and payment, the system reduces wait times, improves efficiency, and enhances the customer experience. The technology is particularly useful in high-traffic retail settings where minimizing employee involvement in routine transactions is desirable. The system may also integrate with inventory management and customer loyalty programs to provide a seamless shopping experience. The automated recognition and payment process ensures secure and efficient transactions while maintaining accuracy and reducing operational costs.
6. The point of sale system as recited in claim 1 , wherein the one or more cameras is a ceiling mounted security camera.
A point of sale (POS) system integrates with a ceiling-mounted security camera to enhance transaction processing and security. The system captures images or video of customers and transactions using the security camera, which is positioned overhead to provide a clear view of the checkout area. The captured visual data is analyzed to verify customer identity, detect fraudulent activities, or monitor compliance with store policies. The system may also use the camera to track customer behavior, such as dwell time or product interactions, to improve operational efficiency. By leveraging existing security infrastructure, the POS system reduces the need for additional hardware while enhancing transaction security and data collection. The integration ensures seamless operation between the POS software and the camera, allowing real-time processing of visual inputs to support various retail functions. This approach improves accuracy in transaction verification and provides additional security layers without requiring dedicated cameras for POS purposes. The system is particularly useful in high-traffic retail environments where security and efficiency are critical.
7. The point of sale system as recited in claim 1 , further programmed to train the feature extractor, the feature generator, and the fully connected classifier with an alternative bi-stage strategy.
A point of sale (POS) system is designed to enhance transaction processing by leveraging machine learning to analyze customer behavior and optimize operations. The system includes a feature extractor, a feature generator, and a fully connected classifier, which work together to process transaction data, generate relevant features, and classify transactions for decision-making. The feature extractor identifies key attributes from raw transaction data, such as purchase history, item categories, and customer demographics. The feature generator transforms these extracted features into a structured format suitable for classification. The fully connected classifier then analyzes the generated features to predict outcomes, such as fraud detection, customer preferences, or inventory management needs. To improve accuracy and efficiency, the system employs an alternative bi-stage training strategy. In the first stage, the feature extractor and feature generator are trained independently using labeled transaction data to refine their ability to identify and transform relevant features. In the second stage, the fully connected classifier is trained using the output from the feature generator, ensuring it learns from the most relevant and optimized features. This bi-stage approach enhances the system's ability to adapt to new transaction patterns and improve decision-making over time. The system is particularly useful in retail environments where real-time transaction analysis is critical for operational efficiency and customer satisfaction.
8. The point of sale system as recited in claim 1 , wherein the feature extractor shares covariance matrices across all classes to transfer intra-class variance from regular classes to the long-tail classes.
This invention relates to a point-of-sale (POS) system that improves classification accuracy for long-tail classes in retail transactions. The system addresses the challenge of imbalanced data in retail environments, where certain items (long-tail classes) appear infrequently, leading to poor classification performance. The system includes a feature extractor that shares covariance matrices across all classes to mitigate this imbalance. By transferring intra-class variance from frequently occurring (regular) classes to the long-tail classes, the feature extractor enhances the system's ability to accurately classify rare items. This approach leverages statistical properties of well-represented classes to improve the representation of underrepresented ones, ensuring consistent performance across the entire product catalog. The system may also include a classifier that processes the extracted features to generate transaction data, further optimizing the POS workflow. The invention is particularly useful in retail settings where inventory diversity is high, and certain items are sold sporadically.
9. The point of sale system as recited in claim 1 , wherein the feature generator optimizes a softmax loss by joint regularization of weights and features through a magnitude of an inner product of the weights and features.
A point-of-sale (POS) system is designed to process transactions efficiently while ensuring security and accuracy. The system includes a feature generator that enhances transaction processing by optimizing a softmax loss function. This optimization is achieved through joint regularization of both weights and features, specifically by controlling the magnitude of the inner product between the weights and features. The regularization process helps improve the system's ability to distinguish between different transaction types or user inputs, reducing errors and enhancing reliability. The feature generator dynamically adjusts the weights and features to minimize the softmax loss, which measures the probability distribution over possible transaction outcomes. By regulating the inner product magnitude, the system ensures that the learned features are both discriminative and robust, improving overall transaction accuracy and security. This approach is particularly useful in environments where transaction data varies widely, such as retail or financial services, where precise classification is critical. The system may also include additional components, such as input interfaces for capturing transaction details and output modules for generating receipts or transaction confirmations, all integrated to streamline the POS workflow.
10. The point of sale system as recited in claim 1 , wherein the feature extractor averages the feature vector with a flipped feature vector, the flipped feature vector being generated from a horizontally flipped frame from one of the plurality of images.
A point of sale (POS) system processes images of items for identification and pricing. The system captures multiple images of an item from different angles or orientations. A feature extractor generates a feature vector representing the item's visual characteristics. To improve robustness against orientation variations, the system generates a flipped feature vector by horizontally flipping one of the captured images and extracting its features. The feature extractor then averages the original feature vector with this flipped feature vector to produce a more stable representation. This averaging helps mitigate discrepancies caused by different item orientations, improving recognition accuracy. The system may further include a database of item profiles, each containing reference feature vectors for comparison. The averaged feature vector is matched against these references to identify the item and retrieve its pricing or other relevant data. This approach enhances the system's ability to handle items presented in various orientations, reducing errors in identification. The system may also include a user interface for displaying recognized items and processing transactions.
11. The point of sale system as recited in claim 1 , wherein each of the plurality of images is selected from the group consisting of an image, a video, and a frame from the video.
A point of sale (POS) system is designed to enhance customer engagement and transaction efficiency by integrating multimedia content into the purchasing process. The system addresses the problem of static, unengaging checkout experiences by dynamically displaying images, videos, or frames extracted from videos during transactions. These visual elements can include product demonstrations, promotional content, or interactive media to improve customer interaction and potentially increase sales. The system captures transaction data and customer behavior while presenting tailored multimedia content, optimizing the checkout process with real-time visual feedback. This approach leverages digital media to create a more immersive and informative shopping experience, distinguishing it from traditional POS systems that rely solely on text or basic graphics. The multimedia content is selected from a predefined set of options, ensuring relevance and alignment with the transaction context. By incorporating dynamic visual elements, the system aims to reduce transaction times, enhance customer satisfaction, and provide retailers with valuable insights into consumer preferences through multimedia engagement metrics.
12. The point of sale system as recited in claim 2 , wherein the communication system connects to a remote server that includes a facial recognition network.
A point of sale (POS) system is designed to enhance transaction security and user authentication by integrating facial recognition technology. The system includes a communication interface that connects to a remote server hosting a facial recognition network. This network processes facial data captured by the POS system to verify the identity of individuals conducting transactions. The facial recognition network compares captured facial features against stored biometric templates to authenticate users, reducing reliance on traditional authentication methods like passwords or PINs. This integration improves transaction security by minimizing fraud risks associated with stolen or lost credentials. The system may also include additional components such as payment processing modules, user interfaces, and data storage for transaction records. By leveraging remote facial recognition, the POS system ensures real-time authentication without requiring extensive local processing, making it suitable for retail, banking, and other transaction-based environments. The technology addresses challenges in secure authentication, particularly in high-volume transaction settings where speed and accuracy are critical. The facial recognition network may also support additional features like user behavior analysis or access control, further enhancing security and operational efficiency.
13. The point of sale system as recited in claim 7 , wherein one stage of the alternative bi-stage strategy fixes the feature extractor and applies the feature generator to generate new transferred features that are more diverse and violate a decision boundary.
A point-of-sale (POS) system incorporates a bi-stage strategy for enhancing transaction security and fraud detection. The system addresses the challenge of improving feature diversity and robustness in machine learning models used for transaction analysis. In one stage of this strategy, the feature extractor remains fixed while the feature generator is applied to produce new transferred features. These features are designed to be more diverse and intentionally violate decision boundaries, thereby improving the model's ability to detect anomalies and fraudulent transactions. The fixed feature extractor ensures consistency in the initial feature representation, while the dynamic feature generator introduces variability to better capture evolving fraud patterns. This approach enhances the system's adaptability and accuracy in identifying suspicious transactions, reducing false positives and negatives. The bi-stage strategy optimizes the balance between stability and adaptability in fraud detection models, making the POS system more resilient against sophisticated fraud attempts. The system leverages machine learning techniques to analyze transaction data, extract relevant features, and apply the bi-stage strategy to improve decision-making processes in real-time transaction processing.
14. The point of sale system as recited in claim 7 , wherein one stage of the alternative bi-stage strategy fixes the fully connected classifier and updates the feature extractor and the feature generator.
A point-of-sale (POS) system integrates machine learning to enhance transaction processing and security. The system addresses challenges in fraud detection and transaction validation by employing a bi-stage strategy for training a neural network. The first stage of this strategy involves fixing a fully connected classifier while dynamically updating both a feature extractor and a feature generator. The feature extractor processes raw transaction data to derive meaningful representations, while the feature generator further refines these representations to improve classification accuracy. By isolating the classifier during this stage, the system ensures that the feature extraction and generation components adapt to new transaction patterns without disrupting the classifier's learned decision boundaries. This approach enhances the system's ability to detect fraudulent transactions while maintaining high accuracy in legitimate transaction approvals. The bi-stage training method allows the system to continuously learn from new data, improving its performance over time without requiring complete retraining of the entire model. This design is particularly useful in high-volume retail environments where transaction speed and security are critical.
15. A computer program product for a point of sale system with facial recognition, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising: receiving, by a processor device, a plurality of images; extracting, by the processor device with a feature extractor utilizing a convolutional neural network (CNN) with an enlarged intra-class variance of long-tail classes, feature vectors for each of the plurality of images; generating, by the processor device with a feature generator, discriminative feature vectors for each of the feature vectors; classifying, by the processor device utilizing a fully connected classifier, an identity from the discriminative feature vector; and controlling an operation of the point of sale system to react in accordance with the identity.
This invention relates to a point-of-sale (POS) system with facial recognition, addressing challenges in accurately identifying individuals, particularly in scenarios with long-tail class distributions where some identities appear infrequently. The system uses a computer program product with a non-transitory storage medium containing executable instructions. The method involves receiving multiple images, extracting feature vectors from each image using a convolutional neural network (CNN) designed to handle long-tail classes by increasing intra-class variance. A feature generator then processes these vectors to produce more discriminative features, which are classified by a fully connected classifier to determine the individual's identity. The system then controls the POS system's operations based on the identified identity, such as authorizing transactions or personalizing services. The CNN's design ensures robust performance even with imbalanced data, improving recognition accuracy in real-world retail environments. The solution integrates seamlessly into existing POS infrastructure, enhancing security and user experience.
16. A computer-implemented method for facial recognition in a point of sale system, the method comprising: receiving, by a processor device, a plurality of images; extracting, by the processor device with a feature extractor utilizing a convolutional neural network (CNN) with an enlarged intra-class variance of long-tail classes, feature vectors for each of the plurality of images; generating, by the processor device with a feature generator, discriminative feature vectors for each of the feature vectors; classifying, by the processor device utilizing a fully connected classifier, an identity from the discriminative feature vector; and controlling an operation of the point of sale system to react in accordance with the identity.
A computer-implemented method for facial recognition in point-of-sale (POS) systems addresses challenges in accurately identifying individuals, particularly in scenarios with long-tail class distributions where some identities appear infrequently. The method involves receiving multiple images of a person and processing them through a convolutional neural network (CNN) designed to handle imbalanced data by enlarging intra-class variance for long-tail classes. This CNN-based feature extractor generates feature vectors for each image, which are then refined into discriminative feature vectors by a feature generator. These refined vectors are classified using a fully connected classifier to determine the individual's identity. The system then controls the POS system's operations based on the recognized identity, such as authorizing transactions or personalizing services. The approach improves recognition accuracy in real-world POS environments where data distribution is often uneven, ensuring reliable identity verification for seamless and secure transactions.
17. The computer-implemented method as recited in claim 16 , wherein controlling includes recognizing a customer and permitting a purchase without an employee intervention.
This invention relates to automated retail systems that enable customer purchases without human intervention. The method involves recognizing a customer, typically through biometric or digital identification, and automatically authorizing a transaction based on pre-stored payment or account information. The system eliminates the need for manual checkout processes, reducing wait times and operational costs. The recognition step may involve facial recognition, fingerprint scanning, or digital wallet authentication, ensuring secure and seamless transactions. The method integrates with existing retail infrastructure, allowing customers to select items and complete purchases autonomously. The system may also track purchase history and preferences to personalize the shopping experience. By automating the checkout process, the invention enhances efficiency in retail environments while maintaining security and convenience for customers. The method is particularly useful in high-traffic stores, self-service kiosks, or automated retail outlets where minimizing human interaction is desirable. The invention ensures accurate identification and payment processing, reducing errors and fraud risks associated with traditional checkout systems.
18. The computer-implemented method as recited in claim 16 , wherein controlling includes logging a customer into the point of sale system and greeting the customer.
A system and method for enhancing customer interaction in a point of sale (POS) environment. The technology addresses the inefficiency and lack of personalization in traditional POS systems, which often require manual customer authentication and lack automated greeting mechanisms. The invention automates customer authentication and greeting processes within a POS system. When a customer approaches the POS, the system identifies the customer, logs them into the system, and generates a personalized greeting. This automation reduces wait times, improves customer experience, and ensures seamless authentication without manual intervention. The system may use biometric data, mobile device proximity, or other identification methods to recognize the customer. Once authenticated, the system retrieves customer preferences, purchase history, or loyalty program details to tailor the greeting and subsequent interactions. The method ensures secure and efficient customer recognition while enhancing engagement through personalized communication. This approach is particularly useful in retail, hospitality, and service industries where quick and personalized customer service is critical. The invention improves operational efficiency by reducing manual steps and enhances customer satisfaction through automated, personalized interactions.
19. The computer-implemented method as recited in claim 16 , wherein controlling includes logging an employee into the point of sale system and greeting the employee.
This invention relates to automated employee authentication and interaction in point-of-sale (POS) systems. The problem addressed is the inefficiency and security risks associated with manual employee login processes in retail environments, where employees must authenticate themselves before accessing the POS system. The solution involves an automated system that detects an employee's presence, authenticates their identity, and seamlessly logs them into the POS system while providing a personalized greeting. The system may use biometric data, RFID badges, or other identification methods to verify the employee's identity. Once authenticated, the system automatically logs the employee into the POS system, eliminating the need for manual entry of credentials. The greeting feature enhances user experience by acknowledging the employee, which may include displaying their name or a custom message on the POS interface. The system may also track login times and employee activity for operational and security purposes. This automation reduces login time, minimizes human error, and strengthens security by ensuring only authorized personnel access the system. The invention is particularly useful in high-traffic retail environments where quick and secure employee authentication is critical.
Unknown
October 6, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.